{"_id":"@aiagentkarl/agent-policy-gateway-mcp","name":"@aiagentkarl/agent-policy-gateway-mcp","dist-tags":{"latest":"0.1.0"},"versions":{"0.1.0":{"name":"@aiagentkarl/agent-policy-gateway-mcp","version":"0.1.0","description":"MCP Server for AI agent policy enforcement - PII detection, guardrails, GDPR/EU AI Act compliance, audit logging, and emergency kill switch","type":"module","bin":{"policy-gateway-server":"index.js"},"main":"index.js","scripts":{"start":"node index.js"},"keywords":["mcp","model-context-protocol","ai-agent","policy","guardrails","pii","gdpr","eu-ai-act","compliance","audit-log","kill-switch","safety"],"author":{"name":"AiAgentKarl"},"license":"MIT","repository":{"type":"git","url":"git+https://github.com/AiAgentKarl/agent-policy-gateway-mcp-js.git"},"dependencies":{"@modelcontextprotocol/sdk":"^1.0.0"},"_id":"@aiagentkarl/agent-policy-gateway-mcp@0.1.0","gitHead":"f467445876b6623bb2e176522075c0c53cced4c9","bugs":{"url":"https://github.com/AiAgentKarl/agent-policy-gateway-mcp-js/issues"},"homepage":"https://github.com/AiAgentKarl/agent-policy-gateway-mcp-js#readme","_nodeVersion":"22.14.0","_npmVersion":"10.9.2","dist":{"integrity":"sha512-OUVgwc5m+1ziQRegVFM4VS7+uAshpQXKG/uDwh5Uv3e1S7z/DyJ72kIdAn3Q2OT0c+9FkDKdPGxW8cd4rMJOvw==","shasum":"df554d8b21c727a70387701f1343594b83f5e8a4","tarball":"https://registry.npmjs.org/@aiagentkarl/agent-policy-gateway-mcp/-/agent-policy-gateway-mcp-0.1.0.tgz","fileCount":3,"unpackedSize":23673,"signatures":[{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEQCICxjMz1+H4Nu62aXHt8WNgjcFUJNOFN+8bn2bCKK49RlAiB1Bb8zaBPDB9+ssC4pthSvEIL9j8uxL7xOVbNn4QGF+w=="}]},"_npmUser":{"name":"aiagentkarl","email":"harrythunder.ai@gmail.com"},"directories":{},"maintainers":[{"name":"aiagentkarl","email":"harrythunder.ai@gmail.com"}],"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/agent-policy-gateway-mcp_0.1.0_1774006982835_0.21041252923776366"},"_hasShrinkwrap":false}},"time":{"created":"2026-03-20T11:43:02.741Z","0.1.0":"2026-03-20T11:43:02.981Z","modified":"2026-03-20T11:43:03.200Z"},"maintainers":[{"name":"aiagentkarl","email":"harrythunder.ai@gmail.com"}],"description":"MCP Server for AI agent policy enforcement - PII detection, guardrails, GDPR/EU AI Act compliance, audit logging, and emergency kill switch","homepage":"https://github.com/AiAgentKarl/agent-policy-gateway-mcp-js#readme","keywords":["mcp","model-context-protocol","ai-agent","policy","guardrails","pii","gdpr","eu-ai-act","compliance","audit-log","kill-switch","safety"],"repository":{"type":"git","url":"git+https://github.com/AiAgentKarl/agent-policy-gateway-mcp-js.git"},"author":{"name":"AiAgentKarl"},"bugs":{"url":"https://github.com/AiAgentKarl/agent-policy-gateway-mcp-js/issues"},"license":"MIT","readme":"# Agent Policy Gateway MCP Server\n\nMCP server for AI agent policy enforcement. Provides PII detection, guardrails, GDPR/EU AI Act compliance checking, audit logging, and an emergency kill switch.\n\n## Features\n\n- **PII Detection** — Scan text for emails, phone numbers, SSNs, credit cards, IBANs with automatic redaction\n- **Guardrails** — Check agent actions against configurable policies (spend limits, domain whitelists, blocked actions)\n- **Audit Logging** — Append-only JSONL audit trail per agent in `~/.agent-audit-log/`\n- **GDPR Compliance** — Check data processing actions against DSGVO requirements\n- **EU AI Act Compliance** — Verify AI operations against EU AI Act risk categories\n- **Emergency Kill Switch** — Instantly block an agent with critical-level audit entry\n\n## Installation\n\n```bash\nnpx @aiagentkarl/agent-policy-gateway-mcp\n```\n\n### Claude Desktop\n\nAdd to your `claude_desktop_config.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"policy-gateway\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"@aiagentkarl/agent-policy-gateway-mcp\"]\n    }\n  }\n}\n```\n\n## Tools\n\n### check_pii\n\nScan text for personally identifiable information using regex patterns.\n\n**Parameters:**\n- `text` (string) — Text to scan\n\n**Returns:** Found PII types, count, masked examples, redacted text, risk level.\n\n### apply_guardrails\n\nCheck an agent action against security policies.\n\n**Parameters:**\n- `action` (string) — Action to check (e.g. \"browse_url\", \"make_purchase\")\n- `context` (object) — Action context with optional `agent_id`, `target_url`, `amount_usd`, `details`\n\n**Returns:** allow/deny/warn decision with violations and recommendations.\n\n### log_action\n\nWrite an entry to the append-only audit log.\n\n**Parameters:**\n- `agent_id` (string) — Unique agent identifier\n- `action` (string) — Action performed\n- `details` (string, optional) — Additional details\n\n### get_audit_log\n\nRetrieve audit log entries for an agent.\n\n**Parameters:**\n- `agent_id` (string) — Agent identifier\n- `limit` (number, optional) — Max entries to return (default: 50)\n\n### check_compliance\n\nCheck GDPR and EU AI Act compliance requirements.\n\n**Parameters:**\n- `action_type` (string) — Type of action (e.g. \"data_collection\", \"profiling\", \"high_risk_ai\", \"agent_operations\")\n- `jurisdiction` (string, optional) — \"gdpr\", \"eu_ai_act\", or \"both\" (default)\n\n**Returns:** Applicable requirements, risk level, and recommendations per framework.\n\n### emergency_stop\n\nKill switch — immediately block an agent from all further actions.\n\n**Parameters:**\n- `agent_id` (string) — Agent to block\n- `reason` (string) — Reason for emergency stop\n\n**Returns:** Confirmation with timestamp. Agent remains blocked until server restart.\n\n## Compliance Coverage\n\n### GDPR / DSGVO\n- Data collection (Art. 6 legal basis, consent)\n- Data processing (Art. 30 records, Art. 32 security)\n- Data sharing (Art. 28 processor agreements)\n- Data deletion (Art. 17 right to erasure)\n- Profiling (Art. 22 automated decision-making)\n\n### EU AI Act\n- High-risk AI systems (Art. 9-15)\n- Limited-risk AI (transparency obligations)\n- General-purpose AI models\n- Prohibited AI practices\n- Agent-specific operations\n\n## Default Guardrail Policies\n\n| Policy | Default Value |\n|--------|--------------|\n| Max spend per action | $100 USD |\n| Allowed domains | github.com, stackoverflow.com, npmjs.com, pypi.org, wikipedia.org, docs.python.org, developer.mozilla.org |\n| Blocked actions | delete_production_data, send_mass_email, modify_permissions, access_admin_panel, execute_raw_sql, disable_security |\n| Require human approval | financial_transaction, data_export, user_deletion, contract_signing |\n\n## Audit Log Format\n\nLogs are stored as JSONL files in `~/.agent-audit-log/{agent_id}.jsonl`:\n\n```json\n{\"timestamp\":\"2026-03-20T10:30:00.000Z\",\"agent_id\":\"agent-1\",\"type\":\"action\",\"action\":\"browse_url\",\"details\":\"...\",\"status\":\"logged\"}\n```\n\n## License\n\nMIT\n","readmeFilename":"README.md","_rev":"1-f8f18521155adb37446ebb11709dce06"}